1 From 702734eebe24adfc91a37f30ed7263b8bfa3361c Mon Sep 17 00:00:00 2001
2 From: Yabin Cui <yabinc@google.com>
3 Date: Fri, 17 May 2019 13:52:31 +0200
4 Subject: perf/ring_buffer: Fix exposing a temporarily decreased data_head
6 [ Upstream commit 1b038c6e05ff70a1e66e3e571c2e6106bdb75f53 ]
8 In perf_output_put_handle(), an IRQ/NMI can happen in below location and
9 write records to the same ring buffer:
12 local_dec_and_test(&rb->nest)
13 ... <-- an IRQ/NMI can happen here
14 rb->user_page->data_head = head;
17 In this case, a value A is written to data_head in the IRQ, then a value
18 B is written to data_head after the IRQ. And A > B. As a result,
19 data_head is temporarily decreased from A to B. And a reader may see
20 data_head < data_tail if it read the buffer frequently enough, which
21 creates unexpected behaviors.
23 This can be fixed by moving dec(&rb->nest) to after updating data_head,
24 which prevents the IRQ/NMI above from updating data_head.
26 [ Split up by peterz. ]
28 Signed-off-by: Yabin Cui <yabinc@google.com>
29 Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
30 Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
31 Cc: Arnaldo Carvalho de Melo <acme@kernel.org>
32 Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
33 Cc: Jiri Olsa <jolsa@redhat.com>
34 Cc: Linus Torvalds <torvalds@linux-foundation.org>
35 Cc: Namhyung Kim <namhyung@kernel.org>
36 Cc: Peter Zijlstra <peterz@infradead.org>
37 Cc: Stephane Eranian <eranian@google.com>
38 Cc: Thomas Gleixner <tglx@linutronix.de>
39 Cc: Vince Weaver <vincent.weaver@maine.edu>
40 Cc: mark.rutland@arm.com
41 Fixes: ef60777c9abd ("perf: Optimize the perf_output() path by removing IRQ-disables")
42 Link: http://lkml.kernel.org/r/20190517115418.224478157@infradead.org
43 Signed-off-by: Ingo Molnar <mingo@kernel.org>
44 Signed-off-by: Sasha Levin <sashal@kernel.org>
46 kernel/events/ring_buffer.c | 24 ++++++++++++++++++++----
47 1 file changed, 20 insertions(+), 4 deletions(-)
49 diff --git a/kernel/events/ring_buffer.c b/kernel/events/ring_buffer.c
50 index 489dc6b60053..fde853270c09 100644
51 --- a/kernel/events/ring_buffer.c
52 +++ b/kernel/events/ring_buffer.c
53 @@ -52,11 +52,18 @@ static void perf_output_put_handle(struct perf_output_handle *handle)
54 head = local_read(&rb->head);
57 - * IRQ/NMI can happen here, which means we can miss a head update.
58 + * IRQ/NMI can happen here and advance @rb->head, causing our
59 + * load above to be stale.
62 - if (!local_dec_and_test(&rb->nest))
64 + * If this isn't the outermost nesting, we don't have to update
65 + * @rb->user_page->data_head.
67 + if (local_read(&rb->nest) > 1) {
68 + local_dec(&rb->nest);
73 * Since the mmap() consumer (userspace) can run on a different CPU:
74 @@ -88,9 +95,18 @@ static void perf_output_put_handle(struct perf_output_handle *handle)
75 rb->user_page->data_head = head;
78 - * Now check if we missed an update -- rely on previous implied
79 - * compiler barriers to force a re-read.
80 + * We must publish the head before decrementing the nest count,
81 + * otherwise an IRQ/NMI can publish a more recent head value and our
82 + * write will (temporarily) publish a stale value.
85 + local_set(&rb->nest, 0);
88 + * Ensure we decrement @rb->nest before we validate the @rb->head.
89 + * Otherwise we cannot be sure we caught the 'last' nested update.
92 if (unlikely(head != local_read(&rb->head))) {